Assessment of discriminatory power of three different fingerprinting methods based on killer toxin sensitivity for the differentiation of Saccharomyces cerevisiae strains P. Buzzini, B. Turchetti and A. Martini Dipartimento di Biologia Vegetale e Biotecnologie Agroambientali, Sezione di Microbiologia Applicata, University of Perugia, Perugia, Italy 2003/0598: received 10 July 2003, revised 14 February 2004 and accepted 15 February 2004 ABSTRACT P. BUZZINI, B. TURCHETTI AND A. MARTINI. 2004. Aims: A panel composed of 44 taxonomically certified strains of Saccharomyces cerevisiae of different origin was used to evaluate the discriminatory power of three different fingerprinting methods based on sensitivity towards 24 killer toxins. Methods and Results: Binary data matrix (BDM), triplet data matrix (TDM) and numerical data matrix (NDM) were used as fingerprinting methods. NDM possessed the highest discriminatory power, assessed through the Simpson’s, and Hunter and Gaston’s indices for the measurement of diversity. The upper limits of fingerprinting ability expressed by the three above methods have been also discussed. Conclusions: NDM determined a significant increase of discriminatory power than the use of BDM or TDM, in terms of an effective amplification of their fingerprinting efficacy. Significance and Impact of the Study: The NDM fingerprinting method could find application in control laboratories for the discrimination of yeast strains of industrial importance or covered by patent. Keywords: binary data matrix, discriminatory power, fingerprinting, killer toxin, triplet data matrix-numerical data matrix, yeast strains. INTRODUCTION Yeast identification conventionally requires a well-defined procedure (up to 90 morphological and physiological tests) (Yarrow 1998), including molecular techniques for the unequivocal assignment of undetermined yeast strains to the species level (i.e. nDNA–nDNA hybridization) (Kurtzman 1998). However, in many instances a further discrimination of individual strains within a given species is needed, especially in patent protection of yeasts possessing desirable industrial properties. Consequently, a number of fingerprinting methods allowing strains belonging to a given species to be discrim- inated with legal certainty, have been developed in recent years (Van der Westhuizen and Pretorius 1992; Niemann et al. 1997; Timmins et al. 1998; Van Looveren et al. 1999; Naumov et al. 2000). The apparently unlimited fingerprinting ability of killer- sensitive relationships, recently demonstrated to be based on different interpretations of strain-related variability (Young 1987; Golubev 1998), has been proposed for fingerprinting yeasts of clinical or industrial interest. Three different data matrices have been used: binary data matrix (BDM), triplet data matrix (TDM) and numerical data matrix (NDM). In BDM presence/absence of sensi- tivity was summarized in a binary sequence of data labelled as Ôkiller sensitivity indexÕ (Buzzini and Martini 2000b; Buzzini and Martini 2000c). In TDM binary data were grouped in triplets and the combined effect within each triplet was expressed by a numerical activity code (Polonelli Correspondence to: Pietro Buzzini, Dipartimento di Biologia Vegetale e Biotecnologie Agroambientali, Sezione di Microbiologia Applicata, Universita`degli Studi di Perugia, Borgo XX Giugno, 06100 Perugia, Italy (e-mail: pbuzzini@unipg.it). ª 2004 The Society for Applied Microbiology Journal of Applied Microbiology 2004, 96, 1194–1201 doi:10.1111/j.1365-2672.2004.02247.x